Recovering Unused Space

You can use ISMF to recover the following kinds of unused space:

Space occupied by members of a partitioned data set that have been updated or deleted
Unused space at the end of a data set
Space occupied by data sets that you no longer need.

Table 1 summarizes the DFSMShsm and DFSMSdss functions that ISMF uses to perform these tasks.

Table 1. Summary of Commands to Recover Unused Space
Task	Function	Scope	Action
Recover unused space at the end of a data set.	CONDENSE	Single data set	Frees unused space at the end of a data set. Compresses partitioned data sets. Performed in the foreground by DFSMShsm.
Recover unused space occupied by members of partitioned data sets.	COMPRESS	Single data set, group of data sets, or volume	Works only with partitioned data sets. Reclaims embedded unused space. Performed in the background by DFSMSdss.
Recover unused space at the end of data sets.	RELEASE	Single data set, group of data sets, or volume	Frees unused space at the end of data sets or frees all allocated, unused space on a specified volume. Performed in the background by DFSMSdss.
Recover space occupied by data sets that are no longer needed.	DELETE /ERASE	Single data set	Deletes or erases online or DFSMShsm-migrated data sets. Performed in the foreground by TSO or DFSMShsm.
Reduce or eliminate free space fragmentation.	DEFRAG	Volume	Relocates non-VSAM and integrated catalog facility cataloged VSAM data set space to reduce free space fragmentation. Performed in the background by DFSMSdss.

Recovering Unused Space from Partitioned Data Sets

There are two ways to reclaim space occupied by members of a partitioned data set that have been updated or deleted. You can use the CONDENSE line operator or the COMPRESS line operator or list command. CONDENSE is performed in the foreground. COMPRESS is performed in the background.

CONDENSE can be used for individual entries in a data set list. The data sets can be either partitioned or sequential. For partitioned data sets, CONDENSE reclaims all embedded unused space. For both partitioned and sequential data sets with secondary allocation, the CONDENSE line operator causes DFSMShsm to freethe unused space at the end of the data set by migrating and recalling the data set. For data sets with secondary allocation, CONDENSE might also reduce the number of extents the data set occupies; when DFSMShsm recalls the data set, it reallocates only the amount of space the data requires. CONDENSE does not change the allocation unit. If the data set was originally allocated in cylinders, some unused tracks might remain at the end of the data set after it is condensed.

On the CONDENSE Entry Panel, you can specify a volume serial number and device type for a specific volume to receive the condensed data set, or you can blank out the field and let DFSMShsm choose the volume. You can also indicate whether you want to wait for the command to execute or return immediately to the list panel. If you choose to wait, ISMF displays the CONDENSE Entry Panel until the CONDENSE is completed. When the data set has been condensed, the list panel is displayed again.

COMPRESS can be used as a line operator or a listcommand on a data set list and as a line operator on a volume list. You can compress a single partitioned data set, a group of partitioned data sets, or a volume containing partitioned data sets. COMPRESS causes DFSMSdss to reclaim all embedded unused space from the specified data sets. However, COMPRESS does not release unused space at the end of a data set. COMPRESS works only with partitioned data sets. If you try to compress a data set that is not partitioned, ISMF displays an error message.

On the COMPRESS Entry Panel, you provide information to control access to the data sets you wish to compress. You can specify the maximum number of times DFSMSdss should attempt to compress the data set and the number of seconds between each attempt. You can also indicate that you want DFSMSdss to use dynamic allocation instead of enqueue to coordinate use of the data set. If the data set is password protected, you must supply the password on this panel.

If you are performing a task against a volume, you can also list the data set names you do not wish to compress.

Recovering Unused Space from Data Sets

Both the CONDENSE line operator and the RELEASE line operator (or list command) free unused space at the end of a data set. When used for a volume, RELEASE frees all allocated, unused space for partitioned and sequential data sets on the volume.

As noted in Recovering Unused Space from Partitioned Data Sets, you can use CONDENSE to free unused space at the end of individual sequential or partitioned data sets. CONDENSE also compresses partitioned data sets.

RELEASE is a DFSMSdss command that you can use against a single data set, a group of sequential or partitioned data sets, or a volume. RELEASE frees space that has been allocated but not used at the end of data sets. Because RELEASE frees only space at the end of data sets, you might want to compress partitioned data sets before releasing the space. Like CONDENSE, RELEASE does not change the allocation unit for a data set. If the data set was originally allocated in cylinders, some unused tracks might remain after the unused space has been released. RELEASE is performed in the background.

On the RELEASE Entry Panel, you can specify a minimum amount of unused space and a minimum secondary allocation that the data set must have for DFSMSdss to execute the command. You can specify the maximum number of times DFSMSdss should attempt to release the unused space and the number of seconds between each attempt. You can also indicate that you want DFSMSdss to use dynamic allocation instead of enqueue to coordinate use of the data set. If the data set is password-protected, you must supply the password on this panel.

If you are performing a task against a volume, you can list names of data sets not to be released.

Deleting or Erasing Data Sets

The DELETE line operator is performed in the foreground by TSO for cataloged data sets, by SVC 29 for uncataloged data sets, and by DFSMShsm for migrated data sets. You can delete individual data sets that have been migrated by DFSMShsm, and data sets that are online. If you enter the DELETE line operator for a data set that has been migrated, ISMF translates the DELETE line operator to an HDELETE and continues processing. You can delete a data set without being certain it is online. You can also delete the DFSMShsm backup versions of a data set.

A TSO message confirms each DELETE. The DELETE line operator is supported for cataloged data sets as well as data sets generated from the VTOC. However, for VSAM data sets in a catalog-generated list, you can issue DELETE only against the cluster entry.

On the DELETE Entry Panel you have the following options:

Delete the data set from the VTOC as well as the catalog.
For VSAM data sets, replace the data with binary.
Delete the data set even if its retention period has not expired.
Delete any or all backup versions. (If you choose to do this, the HBDELETE panel is displayed.)

If the data set is password protected, you must supply the password on the DELETE Entry Panel.

Tip: You can use the ERASE line operator on all application list panels as a synonym for DELETE. The line operator erases single data sets that are online or have been migrated by DFSMShsm.

Defragmenting a Volume

The DEFRAG line operator applies only to volume applications. You can use it to relocate non-VSAM and integrated catalog facility data set extents on a DASD volume to reduce or eliminate free-space fragmentation. When protected data sets or data sets defined with the ERASE option are relocated, information in the old locations is erased for security. DEFRAG is performed in the background by DFSMSdss.

On the DEFRAG Entry Panel, you can specify the number of times DFSMSdss should attempt to gain control of the volume, the number of seconds between each attempt, and the password for protected data sets. You can also specify an index value to control the amount of fragmentation. You can choose among three defragmentation techniques: one minimizes fragmentation, one minimizes data movement, and one minimizes volume involvement. You can list any data sets you wish to exclude from the relocation in the last field on the DEFRAG Entry Panel.